Chemical engineering interviews check whether you can think like someone responsible for a running plant. Expect a few questions on your path and a project you worked on, then solid technical checks on balances, distillation, heat exchangers, pumps, reactors and control loops. Process safety comes up in almost every round, from HAZOP to relief valves to what you do when someone wants to bypass a trip. The rest are plant stories and what-would-you-do scenarios. Each question shows what the interviewer is listening for, a shape for your answer and a short answer you could say out loud. Swap in your own plant, numbers and stories before the day.
Search all questions by round, difficulty and level, or save the ones you want to practise.
Path: the short version, from study to the work you do now.
What hooked you: one specific project or moment, not a general love of science.
Next: the kind of work you want and why this role fits it.
"I picked chemical engineering because I liked chemistry but wanted to work at a scale where things actually get made. What really hooked me was my final-year design project, where we sized a small solvent recovery unit and I saw how many decisions sit behind one line on a flowsheet. After graduating I spent two years as a process engineer on a resin plant, mostly on day-to-day support: checking balances, helping operators through upsets and running small improvement projects. I want to stay close to a running plant, because that's where I learn fastest, but I'd like more ownership now, like leading a unit rather than supporting one. That's why this role appeals to me."
Saying you chose the degree because your marks allowed it, with nothing about the work itself.
What they make: the products and the main process steps, as far as you could find out.
What it means: the engineering problems that kind of plant usually has.
Your fit: where your experience meets those problems.
"From your website and a few trade articles, I understand this site makes specialty surfactants in batch and semi-batch reactors, with a solvent recovery section and its own effluent treatment. That tells me a lot of the engineering here is about reaction control, batch cycle time and getting the most out of your solvents. Those are areas I've worked in. At my last company I spent a year cutting batch cycle time on a sulfonation line, mostly by fixing how we heated up and dosed. I also saw you've been talking publicly about lowering waste per tonne of product, and I'd like to be part of that. I want a site where I can see the whole chain, from raw material to effluent."
Talking only about the company's size or reputation without a word about the process itself.
Problem: what was wrong and why it mattered to the plant.
Your part: what you did yourself, including the analysis.
Result: what changed, and how you know.
Lesson: one thing you would repeat or change.
"At my last plant, steam use on the evaporator section had crept up over two years and nobody knew why. I was given it as my first real project. I started with an energy balance from plant data and found the steam per tonne of water evaporated was well above the design figure. Walking the unit, I found two failed steam traps passing live steam, and a valve left open after an old repair that bypassed the last effect, so we were really running with one effect fewer than the design. I worked with maintenance to replace the traps, and then made the case for the small repair needed to put that effect back in service. Steam use came back close to design within a couple of months. The lesson I kept is to do the balance first, because it tells you where to walk."
Describing what the team did for two minutes without ever saying which part was yours.
Boundary and basis: draw the block diagram, fix the boundary, pick a convenient basis.
Streams and unknowns: list every stream, including small ones, and count unknowns against equations.
Energy and check: choose a reference state, add up the heat terms, then check it closes.
"First I draw a simple block diagram and put a boundary around the part I care about. Then I pick a basis, like one hour of operation or a hundred kilos of feed, so the numbers stay easy. I list every stream crossing the boundary, including the small ones people forget, like vents, purges and drains. For a steady process, total mass in equals total mass out. If there's a reaction, I balance each species using the stoichiometry, or balance on atoms. I count unknowns against independent equations before I start solving. For the energy balance I pick a reference state and account for sensible heat, latent heat, heat of reaction and any heat added or lost. Finally I check that it closes, and if it doesn't, I treat that as a clue, not an annoyance."
Starting to write equations without drawing a boundary or choosing a basis.
Timing and inventory: same period for every reading, tank level changes included.
Meters: calibration, density and temperature compensation, low end of range.
Hidden streams and lab data: vents, drains, flare, samples, open bypasses; analysis quality.
Narrow it: balance smaller sections to find where the gap lives.
"I'd treat the gap as information. First, the time basis: are all the readings over the same period, and did any tank levels change? Inventory changes are one of the most common reasons a balance won't close. Second, the meters. I'd check which ones were calibrated recently, whether flow meters on gases or changing liquids are compensated for density and temperature, and whether any are running near the bottom of their range, where they're least accurate. Third, streams nobody measures, like vents, drains, the flare, sample points or a bypass someone left open. Fourth, the lab data, if I'm balancing a component rather than total mass. A trick I use is to balance around smaller sections, so I can see which piece of the unit the gap lives in, and then go and walk that area."
Forcing the balance to close by adjusting one number without knowing why it was wrong.
The mistake: what you assumed and why it was wrong.
How it was caught: by you or a checker, and what it would have caused.
What changed: the habit you built afterwards.
"Early in my first job, I sized a line for a new transfer between two tanks. I used the viscosity at the design temperature, but the fluid is sometimes moved cold in winter, when it's much thicker. My senior engineer caught it in the check, because he asked what the worst-case temperature was and I didn't have an answer. At the cold condition, the pressure drop was far more than the pump could handle. We went up a pipe size, which was cheap to fix at that stage. I learned two things. One is to ask for the full range of operating conditions, not only the design point. The other is that the check isn't a formality, so now I write my assumptions at the top of every calculation, which makes them easy to challenge."
Claiming you've never made a mistake, or blaming the checker for not catching it sooner.
How it separates: repeated vapour-liquid contact, driven by a difference in volatility.
Raising reflux: usually a purer top product, at the cost of more reboiler and condenser duty.
Limits: minimum reflux, total reflux and flooding.
"A column separates components because they have different volatilities. Vapour from the reboiler rises and meets liquid coming down, on trays or packing. At each contact, the lighter component moves into the vapour and the heavier one into the liquid, so the vapour gets richer in light material as it climbs. At the top, the condenser turns the vapour to liquid, and part of it goes back down as reflux while the rest leaves as distillate. If I raise the reflux ratio on an existing column, I usually get a purer top product, because there's more liquid washing heavy material back down. The cost is energy: more reboiler and condenser duty. And there's a ceiling. Push the internal traffic too far and the column floods, and the separation actually gets worse."
Saying more reflux always means better separation, with no mention of energy cost or flooding.
Duty: from flow, heat capacity and temperature change, or latent heat.
Area: Q equals U times A times the log mean temperature difference, with a correction for multi-pass units.
Layout: which fluid goes tube-side, and a pressure drop check.
Fouling in service: falling duty, falling back-calculated U, rising pressure drop.
"I start from the duty, which comes from the process side: flow times heat capacity times the temperature change, or latent heat if something condenses. Then I use Q equals U times A times the log mean temperature difference, with a correction factor for a multi-pass unit, to get the area. The U value comes from the film coefficients on each side, the wall and a fouling allowance. I also decide which fluid goes in the tubes. Dirty, corrosive or high-pressure fluids usually go tube-side, because tubes are easier to clean and cheaper to make strong. Then I check pressure drop on both sides. Once it's running, fouling shows up as a falling duty, so outlet temperatures drift away from design, and if I back-calculate U from plant data I see it dropping over time. Often the pressure drop creeps up too."
Jumping to area without first working out the duty, or not knowing why a fouling allowance exists.
Definition: how far suction pressure sits above vapour pressure, as head.
Available vs required: the system sets one, the pump curve gives the other; keep a margin.
Symptoms and fixes: noise, unsteady pressure, damage; raise suction head, cut suction losses or cool the liquid.
"NPSH is net positive suction head. The available figure is how far the pressure at the pump suction sits above the liquid's vapour pressure, expressed as head. It depends on the pressure on the liquid surface, the static height, the friction losses in the suction line and the vapour pressure at the pumping temperature. The pump maker gives the required figure on the curve, and I want the available one comfortably above it. If it isn't, the pressure at the impeller eye drops below vapour pressure, bubbles form and then collapse as the pressure rises, which is cavitation. On the plant, you usually hear it first: it sounds like gravel going through the pump. You also see jumpy discharge pressure, falling flow and more vibration. Left alone, it pits the impeller. To fix it, I'd raise the suction level, check for a blocked strainer or cool the liquid."
Mixing up available and required NPSH, or suggesting you control flow by throttling the suction valve.
Concentration: a stirred tank runs at the outlet concentration; plug flow falls along the length.
Volume: for positive-order kinetics, plug flow needs less volume for the same conversion.
When a stirred tank wins: heat control, solids, and selectivity when low reactant concentration helps.
Middle ground: several stirred tanks in series.
"The key difference is concentration. In a well-mixed stirred tank, the whole reactor sits at the outlet concentration, so the reactant is always at its lowest level. In a plug flow reactor, the concentration falls gradually along the length. For a normal positive-order reaction, that means plug flow needs less volume to reach the same conversion. So I'd pick a stirred tank when other things matter more: tight temperature control on an exothermic reaction, solids or a slurry, long residence times in the liquid phase, or when an unwanted side reaction has a higher order in the reactant than the reaction I want, because low reactant concentration then favours my product. For fast gas-phase reactions or high conversion targets I'd lean to plug flow, or several stirred tanks in series as a compromise."
Saying one type is simply better, or not knowing that a stirred tank runs at the outlet concentration.
Arrhenius: the rate constant rises exponentially with temperature.
Heat balance: heat generated rises exponentially, heat removed rises roughly linearly.
Consequence: past the point of no return, the reaction outruns the cooling; control dosing and accumulation.
"The rate constant follows the Arrhenius equation: a pre-exponential factor times e to the minus activation energy over R T. So the rate rises exponentially with temperature, and the higher the activation energy, the more sensitive it is. That matters for an exothermic reaction because the heat it releases rises exponentially with temperature, while the heat the cooling system removes rises roughly in a straight line with the temperature difference to the coolant. Below a certain temperature, cooling wins and a small rise dies away. Above it, heat generation outruns cooling, the temperature climbs, the rate climbs and you get a runaway. That's why we care about cooling capacity and dosing rates, and why we never let unreacted material build up in a reactor before the reaction has started."
Treating rate as rising in a straight line with temperature, or not linking the heat balance to runaway.
Heat removal: jacket area grows slower than volume, so cooling per litre falls; get calorimetry data.
Mixing: longer blend times, hot spots, new side products; decide what to hold constant.
Timing and caution: slow steps get slower; start with a reduced first batch.
"The first thing I'd check is heat removal. If the reactor is ten times the volume, the jacket area grows much less, so the cooling available per litre drops. A reaction that was easy to hold in the pilot could outrun the jacket, so I'd want calorimetry data: the heat of reaction, the adiabatic temperature rise, and what happens if cooling or stirring fails. That usually leads to controlling the reaction by dosing rate rather than trusting the jacket. Second, mixing. Blend times get longer, so hot spots and side products can appear that we never saw at small scale. I'd agree what we hold constant, such as power per volume or tip speed, based on what the reaction is sensitive to. Third, steps like heating up or filtering can take far longer. Once those are answered, I'd still start with a reduced first batch."
Assuming the plant reactor will behave like the pilot one because the chemistry is the same.
Proportional: acts on the current error; usually leaves an offset on its own.
Integral: acts on accumulated error; removes offset, but can oscillate and wind up.
Derivative: acts on rate of change; damps slow loops, amplifies noise, so often left off.
"Proportional action moves the output in proportion to the current error. It reacts straight away, but on its own it usually leaves an offset, so the process settles near the setpoint rather than on it. Integral action adds up the error over time and keeps pushing until the error is gone, which removes that offset. Too much integral makes the loop oscillate, and if the valve is already fully open the integral keeps winding up, so you need anti-windup. Derivative acts on how fast the error is changing, so it can damp overshoot on a slow loop like a reactor temperature. But it amplifies noise, so on fast, noisy loops like flow I'd leave it off. In practice, most loops I've seen on a plant run as PI."
Reciting the three terms with no idea of offset, windup or why derivative is often switched off.
The tag: first letter is the measured variable, the next letters are the function, the number is the loop.
Order of reading: legend sheet, main process flow, then details.
Details that matter: line numbers, valve fail positions, relief devices, isolations and trips.
"Under the usual instrument naming standard, the first letter is what's measured and the letters after it say what the instrument does. So FIC is a flow indicating controller, and 101 is the loop number, which links it to the flow transmitter and control valve in the same loop. On a drawing I've never seen, I start with the legend sheet, because every company tweaks the symbols. Then I trace the main process flow from feed to product before I look at details. After that I check the line numbers, which carry the size, the service and the pipe spec, the fail position of each control valve, where the relief valves are and what they protect, and where the isolation points are. I also look for interlocks and trips, because they show how the unit is meant to fail safely."
Reading symbols one by one without first understanding what the unit is supposed to do.
When: a disturbance hits a faster, measurable variable before it reaches the main one.
Set-up: reactor temperature is the master; its output is the setpoint for a jacket temperature loop.
Rules: inner loop clearly faster; tune inner first; protect the outer loop from windup.
"I use cascade when a disturbance enters through something I can measure faster than my main variable. On a jacketed reactor, what I care about is the reactor temperature, but it responds slowly, and changes in cooling water temperature or pressure hit the jacket first. So I'd make the reactor temperature controller the master, and its output becomes the setpoint for a slave controller on the jacket temperature, which drives the cooling valve. The inner loop catches cooling disturbances before the reactor feels them. For it to work, the inner loop has to be clearly faster than the outer one. When commissioning, I tune the inner loop first on its own, then close the outer loop on top. And I make sure the outer loop doesn't wind up when the inner one is saturated, for example with the cooling valve already fully open."
Tuning the outer loop first, or not knowing that the inner loop must be faster.
Method: nodes, design intent, guidewords applied to parameters.
For each deviation: causes, consequences, existing safeguards, recommendations.
Team and your role: a mixed team with a chair and scribe; you bring the intent, the numbers and follow-up.
"A HAZOP is a structured team review of a design, usually done on the P&IDs. We split the process into nodes, like a feed line or a reactor, and state the design intent for each one. Then the chair applies guidewords, such as no, more, less and reverse, to parameters like flow, pressure, temperature and level. So no flow in the feed line becomes a deviation, and the team works out its causes, the consequences, the safeguards already there, and whether we need a recommendation. A scribe records it all. The team usually includes process, operations, instrumentation, maintenance and safety people. As the process engineer, I bring the design intent and the numbers, like what pressure a blocked outlet would reach, and a clear view of how the process behaves when things go wrong. I also own closing out many of the actions afterwards."
Treating a HAZOP as a checklist to sign off, or listing guidewords without explaining causes, consequences and safeguards.
Scenarios: blocked outlet, fire, control valve failure, tube rupture, loss of cooling, runaway, trapped liquid.
Governing case: calculate each relief load; the one needing the largest orifice sets the size.
Traps: two-phase relief, inlet losses, back pressure, and documenting why cases were ruled out.
"I start by listing every credible way the vessel could be overpressured. Typical ones are a blocked outlet, an external fire, a control valve failing wide open upstream, a tube rupture in a connected exchanger, loss of cooling or reflux, a runaway reaction, and liquid trapped and heated. For each case I work out the relief load, meaning the rate and the phase of what has to leave, at relieving conditions. The case that needs the largest orifice governs the valve size. A few things catch people out. A runaway can relieve as a two-phase mixture, and sizing it as vapour only can badly undersize the valve, so it needs a proper two-phase method. Inlet line losses and back pressure from the flare header can make a valve chatter or cut its capacity. And I document why I ruled cases out, because a reviewer will ask."
Sizing for one obvious case without showing that the others were considered.
Don't approve on the spot: a firing trip is either real or a faulty instrument.
Check: the reading, the setting, and a second measurement.
If a bypass is justified: the site override procedure, extra safeguards, a time limit.
"I wouldn't approve it on the spot. A trip that keeps firing during startup is either telling us something real or the instrument is faulty, and we don't know which yet. So first I'd go and look with the operator: what's the reading, what's the trip setting, and does another measurement agree? If the process really is out of range, the fix is in how we're starting up, not in the trip. If it's a faulty transmitter, then any bypass has to go through the site's override procedure, which normally means a risk assessment, sign-off at the right level, extra safeguards like someone watching a backup reading, and a time limit. I'd rather hold the startup for an hour than bring a unit up with its protection switched off because we felt rushed."
Agreeing to bypass a trip because production is waiting and it is probably a bad instrument.
Name it: a bigger pump is a process change, so it goes through management of change.
Checks: shutoff pressure against design pressure, relief cover, NPSH, motor, materials and seals.
Make it workable: update drawings and procedures, pre-startup check, offer a fast review.
"I'd say I understand the need, but it has to go through management of change, even though it looks like a simple swap. A bigger pump changes the process. I'd check its curve against the system: will the higher shutoff pressure exceed the design pressure of the pipework and equipment downstream, and does the relief protection still cover it? I'd check the suction side has enough NPSH at the new flow, that the motor and power supply are suitable, and that the materials and seals suit the fluid. Then the drawings and operating procedures need updating, and there should be a pre-startup check before it runs. If it's urgent, I'd offer to do the review quickly with the right people, maybe the same day. What I wouldn't do is let it go in over a weekend without anyone looking, because unreviewed changes are a classic cause of incidents."
Treating it as a maintenance job with no process review because the pump fits the same pipework.
What you saw: the specific hazard, in plain words.
What you did: how you stopped the job and who you involved.
Outcome: the fix, the cost, and any change to the system afterwards.
"During a shutdown at my last site, a contractor crew was about to open a flange on a line that had been drained and isolated. Walking past, I noticed the isolation was a single closed valve, not a blind or a double block and bleed, and the line connected to a tank that still had solvent in it. The permit said it was isolated, so on paper they were covered. I asked them to hold, and went to the area supervisor with the drawing. We agreed to fit a spade before breaking the joint. It cost the crew about an hour, and they weren't happy. Afterwards we changed the permit review so the isolation method is written on the permit, not just ticked. What stuck with me is that correct paperwork doesn't mean the job is safe."
Having no example at all, or a story where you noticed something and said nothing.
Stabilise: keep it safe, trim loads to stop making more off-spec.
What changed: feed rate, feed composition, reflux and reboiler duty before the rise.
Narrow it down: loads up points to hydraulic flooding; loads steady points to foaming or tray trouble; use the temperature profile, lab checks and a gamma scan.
"First I'd make sure the unit is safe and ask the operators to stabilise it, usually by trimming the reboiler duty or feed a little, so we stop making more off-spec while we think. Rising pressure drop with worse separation points me towards flooding. So I'd pull the trends: did feed rate, feed composition, reflux or reboiler steam change before the pressure drop started to climb? If the loads went up, it's probably plain hydraulic flooding and backing off fixes it. If the loads didn't change, I'd suspect foaming from a new contaminant in the feed, or fouled or damaged trays. The temperature profile helps show which section is in trouble, and a lab check on the feed can confirm a contaminant. If it looks mechanical, a gamma scan can show where the trouble is without opening the column. Then I'd agree a plan with operations for the morning."
Jumping to damaged trays before checking what changed in the feed and the loads.
Margins: how close each item runs to its limit today, from pumps to columns to the reactor.
Test: a careful step test at higher rate with operations to find the first limit.
Cheap fixes and safety: fix the first bottleneck, find the next, and recheck relief sizing.
"I'd start by listing every piece of equipment in the train and asking how close each one runs to its limit today. For pumps, that's where we sit on the curve and how open the control valve is; for exchangers, the duty against the clean design; for columns, the flooding margin; for the reactor, cooling and residence time. Control valves that already sit nearly wide open are a quick flag. Plant data is the best evidence, so I'd run a short, careful test with operations, raising the rate in steps and watching which limit gets hit first. Then I'd look at cheap fixes for that first bottleneck, like cleaning an exchanger, a bigger impeller or a new control valve, and check what becomes the next limit. I'd also check the relief valves and flare load, because more flow can mean the existing relief sizing no longer covers the worst case."
Only looking at the biggest piece of equipment and forgetting pumps, valves, utilities and relief systems.
Problem and the accepted story: what people believed and why you doubted it.
Your analysis: the data you pulled and the pattern you found.
Fix and proof: what changed and how you know it worked.
"On a crystallisation step at my last plant, batches were failing on filtration time on and off for months. The accepted answer was bad raw material, and we'd rejected a few lots for it. I wasn't convinced, because good and bad batches used the same lots. So I pulled six months of batch records and lined up the slow ones against everything I could find: operator, reactor, time of day, cooling profile. The slow batches were almost all on one reactor, and on those the cooling ran faster than the recipe early in the ramp. The cooling water valve there had a worn seat and passed water even when it should have been nearly shut, so we got fine crystals that clogged the filter. We fixed the valve and added an alarm on cooling-rate deviation, and the failures stopped. I learned to test the popular explanation against data first."
A story with no data, where the cause was found by a hunch that happened to be right.
Locate it: a site solvent balance: bought, used, recovered, lost.
Order of options: avoid at source, then recover, then treat or dispose.
Guard rails: quality sign-off for reused solvent, and track waste per tonne of product.
"I'd start with a solvent balance across the site, because you can't reduce what you haven't located. How much do we buy, where is it used, how much do we recover, and where does the rest go: vents, drums, effluent or product? Usually one or two steps account for most of the loss. Then I'd work in the usual order. Avoid it at source first, for example by cutting wash volumes or reusing the last wash of one batch as the first wash of the next. Then recover it, maybe by improving the recovery column or fixing losses at the condensers. Only then treat or dispose of what's left. I'd check every change with quality, because recovered solvent has to meet the spec for the step it goes back into. And I'd track waste per tonne of product, so growth in production doesn't hide the gain."
Going straight to a new treatment unit without first finding out where the solvent is lost.
Spotting it: what told you energy or yield was being lost.
The fix and its risks: what you proposed and what you checked before doing it.
Proof: how you measured the result afterwards.
"On a distillation unit I supported, the feed went into the column cold and the reboiler used a lot of steam to heat it, while the hot bottoms product was cooled with cooling water before storage. It was a classic heat integration chance. I did the energy balance, showed how much heat we were throwing away, and proposed a feed and bottoms exchanger. Before that I checked there was enough temperature difference, that fouling on the bottoms side was manageable, and that the column would still control well with a preheated feed. It went in during the next shutdown. Steam to the reboiler dropped noticeably, and we also freed up cooling water capacity, which the site was short of. The part I'm proudest of is that I measured the result afterwards rather than quoting the design estimate."
Quoting a big saving with no idea how it was measured.
Situation: the unit and your role on the team.
Problem: what went wrong once real conditions arrived.
Diagnosis and fix: how you found the cause and what you changed.
Lesson: what you now check before every startup.
"I was on the commissioning team for a new evaporator at my last site. During water trials everything looked fine, but when we brought in real product the level control started swinging and we struggled to hold the unit steady. My job was to find out why. I compared the trends with the design data and found two things. The level transmitter was still set up for water density, and the product is much heavier, so the reading was wrong. And the level valve had been sized for design flow, so at our reduced startup rate it was barely open and very touchy. We recalibrated the transmitter for the real density, ran the loop in manual at low rate until we reached design flow, then retuned it. The lesson was to check instrument setup against the real fluid, not just the test fluid, before startup."
A story where everything went to plan, or where the problem was entirely someone else's fault.
The change: what you proposed and why it looked right on paper.
The pushback: what they said, and how you found out the real reason.
Result: the improved plan and how it was adopted.
"At my last plant I wanted to change the startup sequence on a dryer to cut the heat-up time. On paper it was clearly better. The senior operators didn't like it, and at first I thought they were just resisting change. So I asked them to walk me through why. One of them explained that in winter, if you heated the way I proposed, condensate sat in a low point and caused water hammer in the steam line. That wasn't on any drawing. We changed my plan to open a drain first and to ramp the steam more slowly at the start, and we trialled it together on one shift. It still saved a good part of the heat-up time, and because they helped shape it, they were the ones who taught it to the other shifts. Now I talk to the operators before I finalise anything."
Describing operators as the obstacle, or winning by going over their heads.
The issue: the technical problem in one line.
How you explained it: one picture or number, not the theory.
The decision: the options you gave and what they chose.
"We had a reactor that needed an unplanned clean because fouling on the jacket had cut its cooling, which meant longer batches. The planning manager had to decide whether to stop now or run another two weeks. Instead of talking about heat transfer coefficients, I showed one chart: batch time rising week by week, and where it would be in two weeks if the trend held. Then I gave two options in plain terms. Stop now and lose two days, or keep going and lose about the same time through slower batches, with a small risk of a batch going off-spec. They chose to stop at the end of the week. What I learned is that people outside engineering want the decision, the options and the risk, not the physics."
Talking in jargon, or saying non-engineers should just trust the engineer.
Attitude: why time on the plant matters to you.
Experience: any cover or call-outs you've done and what you learned.
Practical question: how the rota and support work here.
"Honestly, it's part of why I want this job. I learn more in an hour walking a unit with an operator than in a day at my desk, and you can't really own a plant from the office. I understand call-outs happen, usually at the worst times, and I'd rather be the engineer who picks up the phone and helps than one who looks at it in the morning. I've done some night cover during a shutdown, and I learned to keep good notes and hand over properly, because the next person needs to know what I changed and why. I'd want to know how the rota works here and what support there is for newer engineers on call, so I can plan around it."
Saying you'd prefer to stay in the office, or agreeing to everything without asking how it works.
Listen first: time on the plant and in the control room, asking what isn't written down.
Close the loop: tell people what happened to what they raised.
Credit and clarity: give maintenance clear priorities and credit the people who spot problems.
"I start from the idea that they see the unit every day, and I mostly see it on my walk-rounds or when something goes wrong. So when I'm new, I spend time in the control room and on the plant asking questions like, what's the most annoying thing about this unit, and what do you do that isn't in the procedure? That tells me where the real problems are. I try to close the loop, too. If someone raises something, I tell them what happened with it, even if the answer is no. With maintenance, I try to give them clear information, like which equipment matters most to the process and why, so they can plan well. And I'm careful to credit the people who spotted a problem. Trust on a plant is built slowly and lost fast."
Seeing yourself as the expert who tells the plant what to do, with nothing about listening.
ClapAssist is an AI interview assistant for Mac and Windows. It listens to the interview on your computer and shows you what to say, in short lines you can read while you talk. Your resume and notes are never stored on our servers. It stays out of screen share on every plan; only you can see it.